CIS AC-VL 2009

8/2/2019 CIS AC-VL 2009

1/18

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [Society for Psychotherapy Research (SPR)]On: 19 October 2009Access details: Access Details: [subscription number 762317397]Publisher RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Psychotherapy ResearchPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713663589

The Collaborative Interactions Scale: A new transcript-based method for theassessment of therapeutic alliance ruptures and resolutions in psychotherapyAntonello Colli a; Vittorio Lingiardi aa Department of Education, University Carlo Bo of Urbino, Urbino & Faculty of Psychology 1, SapienzaUniversity of Rome, Rome, Italy

First Published:November2009

To cite this Article Colli, Antonello and Lingiardi, Vittorio(2009)'The Collaborative Interactions Scale: A new transcript-based method forthe assessment of therapeutic alliance ruptures and resolutions in psychotherapy',Psychotherapy Research,19:6,718 734

To link to this Article: DOI: 10.1080/10503300903121098URL: http://dx.doi.org/10.1080/10503300903121098

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.
http://www.informaworld.com/smpp/title~content=t713663589http://dx.doi.org/10.1080/10503300903121098http://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://dx.doi.org/10.1080/10503300903121098http://www.informaworld.com/smpp/title~content=t713663589

8/2/2019 CIS AC-VL 2009

2/18

The Collaborative Interactions Scale: A new transcript-based method

for the assessment of therapeutic alliance ruptures and resolutions in

psychotherapy

ANTONELLO COLLI & VITTORIO LINGIARDI

Department of Education, University Carlo Bo of Urbino, Urbino & Faculty of Psychology 1, Sapienza University of

Rome, Rome, Italy

(Received 13 July 2008; revised 8 June 2009; accepted 11 June 2009)

AbstractThe authors present a new transcript-based method for the assessment of therapeutic alliance ruptures and resolutions inpsychotherapy*the Collaborative Interaction Scale (CIS)*and discuss the structure and theoretical background of thescale and the rating procedure. To assess interrater reliability, three raters independently evaluated 32 psychotherapysessions (2,984 patient utterances and 2,984 therapist utterances) using the CIS, which demonstrated good interrater

reliability (average k0.66.81). In evaluating the relationship between therapist interventions and patient alliance ruptureand collaborative processes, the authors found significant correlations between therapist negative interventions and patientalliance ruptures and among therapist positive interventions, patient collaborative processes, and indirect rupture markers.Results indicate that the CIS is a reliable rating system, useful in both empirical research and clinical assessments.

Keywords: therapeutic alliance; alliance ruptures; alliance resolutions; collaborative interaction scale; assessment

The concept of therapeutic alliance emerged histori-

cally from psychodynamic literature (Freud, 1912/

1958; Greenson, 1965; Zetzel, 1956) but has been

found to be crucial in various psychotherapeutic

approaches. Much of the original research on

therapeutic alliance focused on providing empiricalevidence of its relationship to the outcome of

psychotherapy. Although research findings show

that a strong positive therapeutic alliance is quite

consistently related to a positive outcome (Horvath

& Bedi, 2002; Martin, Garske, & Davis, 2000;

Wampold, 2001), the exact nature of this relation-

ship needs further investigation.

Some authors suggest that therapeutic alliance is

an active agent of the psychotherapy change process

(Bordin, 1979; Horvath, 1994) and that shifts in the

collaboration levels can be considered fundamental

change keys (Safran, Crocker, McMain, & Murray,

1990; Safran & Muran, 2000a). These fluctuationswere conceptualized in various terms: strains in the

alliance (Bordin, 1994), weakenings and repairs of

the alliance (Lansford, 1986), impasses in the

therapeutic relationship (Elkind, 1992), and ther-

apeutic alliance ruptures and repairs (or resolutions;

Safran & Muran, 2000a).

Several studies have confirmed the existence of a

rupturerepair pattern and its relationship to im-

proved therapy outcome (Kivlighan & Shaughnessy,

2000; Stiles et al., 2004; Strauss et al., 2006). Other

research found local rupturerepair patterns in 50%

of the cases but no relationship with outcome(Stevens, Muran, Safran, Gorman, & Winston,

2007).

Some studies compared the efficacy of treatments

focused on alliance rupture and resolution with that

of other forms of intervention. One study compar-

ing a specific alliance-focused intervention, brief

relational therapy, with short-term psychodynamic

therapy and cognitivebehavioral therapy found

significant differences in two areas: Brief relational

and cognitive-behavioral models produced more

clinically significant change while brief relational

therapy had significantly lower dropout rates

(Muran, Safran, Samstag, & Winston, 2005). Otherresearchers evaluated the efficacy of an integrative

form of cognitive therapy (ICT) for depression

that incorporates specific strategies for addressing

alliance ruptures and compared it with a traditional

cognitive therapy (CT): Effect size estimates

revealed that ICT patients evidenced greater

Correspondence concerning this article should be addressed to Vittorio Lingiardi, Department of Dynamic and Clinical Psychology,

University of Rome, Via Dei Marsi, 78, Rome 00185, Italy. E-mail: [email protected]

Psychotherapy Research, November 2009; 19(6): 718734

ISSN 1050-3307 print/ISSN 1468-4381 online # 2009 Society for Psychotherapy Research

DOI: 10.1080/10503300903121098

8/2/2019 CIS AC-VL 2009

3/18

posttreatment improvement and more clinically

significant change than CT patients (Constantino

et al., 2008). Moreover, a number of studies

indicate that alliance improves when therapists are

able to focus on the relationship and address

ruptures nondefensively (for a review, see Safran

et al., 2002). In a recent study, based on a sample

of 128 patients randomly assigned to three different

time-limited psychotherapies for personality disor-ders (cognitive-behavioral, brief relational and

short-term dynamic) results indicate that lower

rupture intensity and higher rupture resolution are

associated with better ratings of the alliance and

session quality, and lower rupture intensity predicts

good outcome on measures of interpersonal func-

tioning, while higher rupture resolution predicts

better retention (Muran, Safran, Gorman, Samstag,

Eubanks-Carter, & Winston, 2009). Finally, based

on available empirical evidence, the repair of

alliance ruptures has been included in a list of

promising and probably effective treatment princi-

ples (Norcross, 2002).The history of alliance research is linked to the

development of many different measures for its

assessment (Elvins & Green, 2008; Horvath &

Bedi, 2002), such as the Penn Helping Alliance

Scales (Alexander & Luborsky, 1987; Luborsky,

CritsChristoph, Alexander, Margolis, & Cohen,

1983), the Vanderbilt Scales (Suh, Strupp, &

OMalley, 1986), the Working Alliance Inventory

(WAI; Horvath, 1982; Horvath & Greenberg,

1989), and the California Alliance Scales (CALPAS;

Gaston & Marmar, 1994). These tools have demon-

strated their reliability, with fair to good convergent

validity (Tichenor & Hill, 1989).

Although we have a proliferation of instruments

to assess therapeutic alliance, there is a shortage of

measures assessing in-session therapeutic alliance

fluctuation. Only two scales were specifically devel-

oped for this purpose: the Rupture and Resolution

Scale (RRS; Samstag, Safran, & Muran, 2000;

Samstag, Safran, Muran, & Stevens, 2002) and

the Menninger Alliance Rating Scale (MARS; Allen,

Newsom, Gabbard, & Coyne, 1984). Although both

of these measures proved quite reliable (Allen

et al.,1984; Samstag et al., 2002), they have not

been further validated. Moreover, the MARS, atranscript-based measure, only furnishes informa-

tion regarding shifts in patient collaboration. It does

not describe typologies of collaboration/noncolla-

boration, and therapist interventions are evaluated

only considering their form (e.g., confrontation,

interpretation) and focus (therapeutic relationship

or not). The RRS does not specifically address

therapist interventions that contribute negatively to

the psychotherapy process, nor does it quantify the

level of collaboration of patient and therapist.

Finally, as described in its coding manual (Samstag

et al., 2000), RRS needs both transcribed and

videotaped sessions.

A number of studies used self-report alliance

measures to assess alliance fluctuations and alliance

ruptures (Kivlighan & Shaughnessy, 2000; Stiles et

al., 2004; Strauss et al., 2006; Stevens et al., 2007).However, as has been pointed out (Westen &

Shedler, 1999a, 1999b), these kinds of measures

can be faulty (a) because of poor self-reflection or

any type of bias on either the patients or the

therapists part and (b) because they use a retro-

spective (postsession) recollection of the session. For

example, using self-reports, patients could not re-

member the effort made by the therapist to over-

come a rupture because they were in an angry state

after the session; other patients could acknowledge

with difficulty the idea that the therapist did his or

her job in creating a negative atmosphere; others

might simply dissociate emotionally marked rela-tional episodes after the session.

Traditionally, therapeutic alliance measures such

as WAI and CALPAS (including in the observer

versions), which evaluate the therapeutic alliance at a

macrolevel, seem more suitable for assessing ther-

apeutic alliance as a general factor related to the

outcome than for depicting the idiosyncratic inter-

actional patterns that unfold between patient and

therapist (Charmann, 2004, p. 18). Because these

measures can only study shifts between sessions

rather than within the session itself, some rupture

events may go undetected (Stevens et al., 2007). In

short, these methodologies described shifts in

alliance but did not directly examine in-session

transactions. We can only infer that ruptures were

captured by our quantitative method (Strauss et al.,

2006, p. 344).

Studying in-session transactions through a micro-

analytic investigation of the way patient and therapist

coconstruct their alliance could be a way to detect

clinically useful guidance regarding the therapeutic

relationship (Horvath, 2006). Such indicators could

increase our knowledge of how interventions can

affect therapeutic alliance (Ackerman & Hilsenroth,

2001, 2003).In this study, we present a new transcript-based

method to assess therapeutic alliance ruptures and

resolutions in psychotherapy along with preliminary

data on its reliability. Before describing scale struc-

ture, rating procedure, and research results, we

present the theoretical and empirical framework on

which we based the scale construction and develop-

ment procedure.

The Collaborative Interactions Scale 719

8/2/2019 CIS AC-VL 2009

4/18

Theoretical and Empirical Framework

Alliance Ruptures and Two-Person Psychology. Alli-

ance rupture is a very slippery concept (Safran &

Muran, 2006, p. 288). An alliance rupture has been

defined in various ways: as an impairment or

fluctuation in the quality of the alliance between

therapist and client (Safran et al., 1990), a tension or

breakdown in the collaborative relationship between

patient and therapist (Safran et al., 2002), a dete-

rioration in the relationship (Safran & Muran,

2000b), a problem in the quality of relatedness

or a deterioration in the communicative process

(Safran & Muran, 2006). The term has also been

used to indicate severe disruptions in the therapeutic

alliance or even momentary and subtle fluctuations

in the quality of the therapeutic relationship (Safran

et al., 1990). In our opinion, a particularly clear and

useful definition of the construct identifies the

alliance rupture as an impairment or fluctuations

in the quality of the alliance between the therapist

and client (Safran et al., 1990, p. 154). The qualityof the therapeutic alliance can be defined as a

function of the degree of agreement between thera-

pist and client about the goals and tasks of psy-

chotherapy that is mediated by the quality of the

relational bond between therapist and patient

(Safran et al., 1990, p. 154).

The concept of alliance rupture and repair implies

a conceptualization of the therapeutic alliance as an

important dimension of the therapeutic relationship

that involves an on-going process of intersubjective

negotiation (Safran & Muran, 2000a, p. 165). This

highlights the fact that the alliance is not a static

variable that is necessary for the therapeutic inter-vention to work but rather a constantly shifting,

emergent property of the therapeutic relationship

(Safran & Muran, 2006, p. 288). This kind of

conceptualization was already present in the work

of Bordin (1994), who, in contrast to traditional

psychoanalytic authors, emphasized the negotiation

of task and goals as important steps in alliance

building and attaining the strength to overcome

strains and ruptures (p. 15).

Safran and Muran (2000a) also provided an

empirically derived process model of therapeutic

alliance negotiation in which alliance ruptures arecodetermined by therapist and patient. This model is

based on a definition of psychotherapy processes and

therapeutic impasses and resistances in light of a

two-person psychology. From this point of view, any

apparent obstruction in the therapeutic process

must be understood as a function of the interaction

between the patient and the therapist: For example,

a patient who has difficulty accessing painful

emotional material is having difficulty accessing it

in a specific relational context (Safran & Muran,

2000a, p. 80). From a relational and interpersonal

perspective, resistance is interpreted not only as a

client character issue but also as the product (at

conscious and unconscious levels) of the interperso-

nal matrix in which it is produced. Similarly,

therapist negative contributions to the relationship

are interpreted not only as a therapist problem but

also as the product of the interpersonal matrix inwhich they are produced.

More generally, it should be emphasized that

therapeutic alliance ruptures (of patient and thera-

pist) are inevitable aspects of the therapeutic process.

From an intersubjective perspective, the process of

ruptures and repairs characterizes all relationships

and thus the therapeutic relationship as well (Beebe

& Lachmann, 2002). Although this process can even

assume the form of severe disruptions in the relation-

ship, it must be considered as an unavoidable and

natural feature of the relationship. Of course, the

features of this process are influenced by, for

example, patient and therapist personality organiza-tions, relational patterns, and interpersonal

schemata.

Patient Contribution to Alliance Ruptures and

Resolutions. Safran and Muran (2000a), in accor-

dance with Harpers work, have organized patient

ruptures into two main subtypes: withdrawal and

confrontation (Harper, 1989a, 1989b). In withdra-

wal ruptures, the patient withdraws or partially

disengages from the therapist, his or her own

emotions, or some aspect of the therapeutic process

(Safran & Muran, 2000a, p. 141). Withdrawalmarkers include patient behaviors such as verbal

disengagement (e.g., changing topic, long silences,

or use of vague, abstract language) or a mismatch

between affective expression and narrative content.

In this kind of marker, the patient indirectly

expresses disaffection or disagreement about the

tasks or goals of therapy or about the relationship.

Withdrawal markers also include patient avoid-

ance maneuvers, such as skipping from topic to topic

in such a way as to prevent therapist interventions in

order to reduce patient anxiety associated with a

rupture in the alliance. Another maneuver involves

self-esteem-enhancing operations, in which the cli-ent may attempt to justify or defend him- or herself

during the process of a rupture as a means of

boosting a deflated sense of self-worth. As observed

by Safran et al. (1990), avoidance maneuvers and

self-esteem-enhancing operations can be considered

as reflections of what Sullivan (1953) termed

security operations. Conversely, in confrontation

ruptures, the patient directly expresses anger,

resentment, or disaffection with the therapist or

720 A. Colli and V. Lingiardi

8/2/2019 CIS AC-VL 2009

5/18

some aspect of the therapy (Safran & Muran,

2000a, p. 141). Examples of this kind of marker

are most evident in the patients verbal criticisms

of the therapist, either as a person or in terms of his

or her professional qualifications. These generally

appear as hostile or dismissive manners of

communication.

Although Safran and Muran (2000a) highlighted

and explicitly formalized the role of the therapist inthe process of addressing ruptures, they did not view

the resolution process as a one-sided enterprise: It

invariably includes the patient willingness to partici-

pate in a process of collaborative inquiry about the

nature of the interactive matrix. In some way the

patient must also be willing and able to step out of

the enactment in order to begin an exploration of

what is going on in the therapeutic relationship

(Safran & Muran, 2000a, p. 145). Other authors

define patient collaboration as the extent to which

the patient is bringing in significant issues and

making good use of the therapists efforts (Allen

et al., 1984) or as the patients capacity to self-disclose intimate and salient information, to self-

observe ones reactions, to explore contributions to

problems, to experience emotions in a modulated

fashion, to work actively with the therapists com-

ments, to deepen the exploration of salient themes

(Gaston & Marmar, 1994, p. 89). These definitions

share the importance of patient expressions of

feelings and thought, of significant issues, and of

reflections or self-observations of his or her conflicts

and feelings.

Therapist Contribution to Alliance Ruptures andResolutions. As Strupp pointed out, Major decre-

ments to the foundation of a good working alliance

are not only the patients character distortions and

maladaptive defenses but*at least equally

important*the therapists personal reactions

(Strupp, 1980, p. 953). For example, the therapist

may explore patient negative feelings, such as

hostility or flirting, in a collaborative way but can

also respond to these feelings in a complementary

way or avoid their exploration entirely by shifting the

focus of investigation. The therapists contribution

to the therapeutic alliance may be divided into two

main discernible but interdependent components:relational (e.g., empathy, attunement, warmth) and

technical (e.g., type of intervention, focus of the

intervention).

Several studies suggest that key elements of

empathy (such as warmth/friendliness, affirming,

helping, and understanding) are positively associated

with therapeutic alliance (Ackerman & Hilsenroth,

2001). Conversely, there is a consensus among

studies that a poor alliance is related to therapists

who are not confident in their ability, tired, rigid,

critical, distant, bored, defensive, or blaming (Acker-

man & Hilsenroth, 2001), and that these character-

istics evoke more hostile resistances in patients

(Marmar, Weiss & Gaston, 1989). Such negative

therapist states were also present in therapist beha-

viors related to ruptures in the alliance (Ackerman &

Hilsenroth, 2001). These results have been con-

firmed by other research using structural analysis ofsocial behavior (Benjamin, 1984). These studies

demonstrated that successful outcome cases are

characterized by a high proportion of therapist

statements that express understanding, attentive

listening, and receptive openness (Henry, Schact,

& Strupp, 1986; Watson, Enright, & Kalogerakos,

1998). Critical, hostile, and controlling statements,

on the other hand, are negatively associated with

good outcomes (Henry et al., 1986; Watson et al.,

1998).

The strategic interventions used by the therapist

may be considered another component of the overall

alliance. In a review on the relationship betweentherapist interventions and therapeutic alliance,

several therapist techniques (e.g., exploration, depth,

accurate interpretation, facilitating expression of

affect, reflection, attending to patients experience,

supporting) were found to contribute positively to

the therapeutic alliance (Ackerman & Hilsenroth,

2003).

As observed by Ackerman and Hilsenroth (2003),

It is interesting to note that the . . . therapists

significant contributions to the development and

maintenance of the alliance are similar to the

features identified as useful in the identification

and repair of ruptures in the alliance (p. 29). These

interventions were identified and categorized by

Safran and Muran (2000a) in general principles of

intervention and specific principles of communica-

tion. Some of the general principles are to (a)

establish a sense of we-ness, (b) emphasize ones

own subjectivity, and (c) focus on the here and now.

Some of the specific communication principles are to

(a) disclose experience or acknowledge ones own

actions and (b) provide feedback regarding subjec-

tive experience or perceptions of patient. These

principles may help therapist and patient resolve a

therapeutic impasse, stepping outside of the negativerelational cycle that is occurring. For example, by

establishing a sense of we-ness and framing the

impasse as a shared experience, the therapist begins

transforming a rupture into a collaborative effort; for

example, focusing on concrete and specific aspects

instead of on general and abstract instances pro-

motes experiential awareness.

Therapists may also contribute in a negative way

to the psychotherapy process through misapplication


8/2/2019 CIS AC-VL 2009

6/18

of the techniques. Misapplication can take the form

of unyielding attempts to link a patients inappropri-

ate reactions toward the therapist to earlier con-

flicted relationships with parental figures (Marmar

et al., 1989), inflexibility and destructive behaviors

(Eaton et al., 1993), or a lack of responsiveness to

explore the patients feelings (Piper et al., 1999).

Several studies have investigated therapist behaviors

in cases of unresolved misunderstandings: forexample, inflexible adherence to strategies, being

hypercritical or unresponsive, giving unwanted

advice.

From a different point of view, the therapist degree

of collaboration may be defined as the quality of his

or her use of previous patient communications.

Waldron et al. (2004) consider several variables to

evaluate therapist activities (e.g., type of interven-

tion, focus of the intervention). What is important

for our discussion here is that the quality of the

intervention is evaluated as the degree to which the

therapist follows the patients immediate emotional

focus, intervenes in a way that shows psychologicalcontinuity with previous contributions, is apt in

content, timely and tactful, and comments in a way

that makes psychological appeal to the patient

(Waldron et al., 2004, p. 449).

In summary, evaluation of therapist activities must

take in consideration at least two aspects: the quality

of the intervention (timing, attunement, tactfulness,

comprehensibility) and the form of the intervention

(e.g., clarification, confrontation, interpretation).

Scale Development and Item Derivations

The Collaborative Interaction Scale (CIS) is the

result of 8 years of ongoing research and transcript-

based investigation of therapeutic alliance and

rupturerepair processes. The first version of the

scale, formerly known in Italian as IVAT (Indice di

Valutazione dellAlleanza Terapeutica [Therapeutic

Alliance Evaluation Index]), was presented at the

2001 conference of the Italian Society for Psy-

chotherapy Research (Colli & Lingiardi, 2001).

The scale has been revised and tested in several

pilot studies (Colli & Lingiardi, 2002, 2003, 2005,

2006). Initially, the structure of the scale was defined

and then single items were selected.In light of the theoretical and empirical findings

reported in the previous section, we structured the

scale into two main scales: one for the evaluation of

patient contributions to the process (CIS-P) and one

for therapist contributions (CIS-T), each with sub-

scales. The CIS-P is composed of three subscales

evaluating patients positive and negative contribu-

tions: the Collaborative Processes scale, the Direct

Rupture Markers scale, and the Indirect Rupture

Markers scale. Similarly, the CIS-T is composed of

two subscales evaluating therapists contributions to

the psychotherapeutic process: the Positive Interven-

tions scale and the Negative Interventions scale. We

labeled the two main categories of patient ruptures

direct and indirect (and not withdrawal and

confrontation as proposed by Safran and Muran)

because we wanted to stress their descriptive and

easily recognizable dimension. We give a moredetailed description of the scale in the next section.

Items and rating criteria have been derived from

several sources: empirical and clinical literature on

alliance ruptures, measures of therapeutic alliance,

qualitative evaluations of psychotherapy session

transcripts, and comments and suggestions of expert

clinicians. The list of rupture markers has been

derived in large part from Safran and Muran studies

(Safran et al., 1990; Safran, Muran, & Samstag,

1994), in particular from a list of behaviors asso-

ciated with rupture states contained in Samstag,

Safran, and Muran (2004).

Patient collaborative processes levels were inspiredby the CALPAS definition of patient working

capacity as the patients capacity to self-disclose

intimate and salient information, to self observe

ones reactions, to explore contributions to pro-

blems, to experience emotions in a modulated

fashion, to work actively with the therapists com-

ments, to deepen the exploration of salient themes

(Gaston & Marmar, 1994, p. 89). The concept of

collaborative process was organized according to

Horowitz et al.s (1993) categories of patient ela-

boration (convey facts, convey emotionality, convey

significance). A large part of therapist positive

interventions were derived from Safran and Murans

(2000a) specific principles of intervention and their

related clinical examples. These have been included

because they prove to be particularly useful in

managing ruptures and building therapeutic alliance.

Negative therapist interventions were inspired by

the Vanderbilt Negative Indicator Scale (Suh,

Strupp, & OMalley 1986). Others were formulated

reversing Safran and Murans principles of inter-

ventions (2000a), for example, a therapist who

distances from the emotional content by minimizing,

intellectualizing, or talking in a technical jargon

instead of exploring patients feelings and focusingon his or her concrete experience.

We obtained a preliminary set of 56 items. Then we

asked a pool of expert clinicians to code the patient

rupture items into two major rupture types: direct

and indirect (Safran & Muran, 2000a). Using this

preliminary set of items, five expert clinicians (two

cognitive and three dynamic) coded 16 transcripts

each from 16 different patients. On the basis of this

first application, and considering item reliability, we


8/2/2019 CIS AC-VL 2009

7/18

prepared a second version of the scale. This was

tested on a sample of 65 junior psychotherapists, who

coded two transcripts of two different patients (one

session per patient). Finally, we prepared a third

version of the scale and a new coding manual. This

last version was tested by three raters, who coded six

sessions each of three different patients.

At the end of this process, we selected the most

salient and reliable items to obtain the present 41-item version of the scale. A pilot study with three

experienced raters showed a good reliability of the

present version of the scale, with mean k values of

.85 for direct rupture markers (DRMs), .71 for

indirect rupture markers (IRMs), .73 for collabora-

tive processes (CPs), .81 for positive interventions

(PIs), and .74 for negative interventions (NIs; Colli

& Lingiardi, 2007).

Collaborative Interaction Scale

The CIS is a rating system for the assessment of

alliance ruptures and repairs in psychotherapy (for acomplete item list, see Table I). External raters

conduct their evaluation on transcripts. The coding

procedure and manual are tailored for transcript

evaluations, but audiotapes, if available, can be very

useful for a more detailed evaluation. The CIS is

composed of two main scales: one for the evaluation

of patient rupture and collaborative processes, CIS-

P, and one for the evaluation of therapist positive

and negative contributions to the therapeutic rela-

tionship, CIS-T.

The CIS-P Direct Rupture Markers scale com-

prises nine items; the Indirect Rupture Markers

Scale, nine items; and the Collaborative Processes

Scale, three items. DRMs are characterized by an

aggressive and accusatory statement of resentment

or dissatisfaction in regard to the therapist or some

aspect of the therapy process (Safran et al., 2008).

Examples of DRMs are when the patient acts in a

hostile or sarcastic manner, complains heatedly

about lack of progress, questions or rejects the tasks

or the goals of therapy, and so on. Examples of IRMs

are when the patient indirectly expresses a form of

emotional disengagement from the therapist, from

some aspect of the therapy process, or from his or

her internal experience (Safran et al., 2008). Forexample, the patient may skip from topic to topic in

a manner that prevents the therapist from exploring

the issue in depth, may respond in an overly

intellectualized fashion, or may be overcompliant

or submissive.

Patient CPs include when the patient brings

salient and significant themes, shares intimate and

salient information with the therapist, self-observes

his or her reactions, or works actively with the

therapists comments. The patient could give a

positive contribution to the psychotherapeutic pro-

cess in various ways: speaking of new and significant

facts, reflecting on personal feelings, having an

insight about his or her history, relationships, and

so on.

Further descriptions and exemplifications of

DRMs, IRMs, and CPs are provided in the coding

manual and also reported next in the RatingProcedure section. In conclusion, we wish to stress

that when a patient communicates in a collaborative

way that he or she disagrees with or does not

understand a therapist comment, this clearly must

not be considered a rupture marker. Table II shows

examples of different patient responses to and

codings for the same therapist intervention.

The CIS-T Positive Intervention scale is com-

posed of 12 items that evaluate collaborative and

repairing therapist interventions. The eight-item

CIS-T Negative Intervention scale evaluates nega-

tive contributions of the therapist. We define a PI as

a therapist intervention that, in relation to previouspatient communications, is emotionally attuned,

focused on patient experience, and linguistically

clear. Quality of the intervention is a necesssary

condition for rating it as positive. We do not consider

a priori an intervention as positive simply because of

its form. To be evaluated as positive, a therapist

intervention may take various forms, such as when

the therapist focuses attention on the here and now

of the relationship, reframes tasks/goals of therapy,

self-discloses countertransference feelings, and so

on. We define an NI as a therapist intervention that

is not emotionally attuned, not focused on patient

concrete experience, or not linguistically clear.Therapist NI can assume several forms, such as

when the therapist shows hostility, talks in technical

jargon, seems to impose his or her worldview

about ethical problems, is critical, and so on.

Although the CIS is rooted in the psychodynamic

relational and cognitiveinterpersonal approaches,

its items are written in a transtheoretical language,

which makes it useful for researchers from a variety

of backgrounds.

Rating Procedure. The basic evaluation unit of the

CIS is the individual speaking turn of eitherthe patient or the therapist. The rater assesses the

presence, in each patient utterance, of a DRM,

IRM, or CP and, in each therapist utterance, of a

PI or NI. Each patient and therapist utterance is

also assessed according to a collaboration/noncolla-

boration score. Rupture intensity and collaboration

level intensity are evaluated on a 3-point Likert

scale (for ruptures: 10low rupture intensity; 20

mid rupture intensity; 30high rupture intensity;


8/2/2019 CIS AC-VL 2009

8/18

Table I. Mean Interrater Reliability (Kappa), Presence, and Intensity Values for Each Item

Item ka P M SD

Direct rupture markers

DRM1: Patient doesnt agree with therapist about therapy tasks or goals .75 4.33 2.92 0.41

DRM2: Patient criticizes therapist as a person or for his/her competence .89 10.33 2.92 0.35

DRM3: Patient strongly refuses a therapist inter vention or feels uncomfor table .84 15. 13 2.87 0. 61

DRM4: Patient complains about lack of progress .88 4.77 2.55 0.71

DRM5: Patient doubts about current session .71 5.33 2.87 0.62

DRM6: Patient doubts about being in therapy .79 4.33 2.61 0.45DRM7: Patient complains about parameters of therapy (e.g., session time, fee) .92 5.66 2.71 0. 44

DRM8: Patient doubts about feeling better .86 4.33 2.63 0.45

DRM9: Patient is sarcastic toward therapist .68 3.66 2.51 0.35

DRM .81 3.33 2.91 0.35

Indirect rupture markers

IRM1: Patient talks in wordy manner and/or spends inordinate amount of time talking

about other people and their doings or overly elaborates nonsignificant stories and so on

.69 36.33 1.06 0.26

IRM2: Patient changes topic or tangentially answers to therapist inter vention .72 87. 00 1.33 0. 45

IRM3: Patient short answers to therapist open question .75 89.00 1.66 0.45

IRM4: Patient denies evident feeling state (e.g., anger, fear, shame) .61 32.33 1.13 0.24

IRM5: Patient intellectualizes about his/her inner experience .62 115.25 1.25 0.27

IRM6: Patient alludes to negative sentiments or concerns about therapeutic relationship through

a thematically linked discussion of out-of-session events or relationships

.67 6.33 1.66 0.51

IRM7: Patient interacts in a acquiescent manner .57 23.00 1.13 0.54

IRM8: Patient uses self-enhancing strategies or self-justifying statements .69 51.00 1.11 0.46

IRM9: Patient is self-critical or self-blaming .69 67.33 1.13 0.55

IRM .66 69.66 1.07 0.11

Collaborative processes

CP1: Patient talks about new significant fact, introduces a topic or elements within a topic .69 1578.33 1.01 0.31

CP2: Patient talks about his/her feeling and/or thoughts, makes clear intensity

or quality of his/her feelings or attitude

.73 654.33 1.92 0.42

CP3: Patient talks about meaning of events or connects topic to a topic or to a schema, etc. .76 34.66 2.87 0.33

CP .72 171.33 1.01 0.31

Positive interventions

PI1: Therapist focuses on the here and now of the relationship .66 43.22 2.87 0.34

PI2: Therapist explores different patient states .65 234.66 1.11 0.33

PI3: Therapist provides a feedback to the patient .62 91.66 1.13 0.34

PI4: Therapist suggests a patient emotion .61 236.13 1.13 0.31PI5: Therapist believes that patient is indirectly talking about relationship .69 6.33 2.83 0.27

PI6: Therapist furnishes an empathic sustain to patient .63 82.00 2.11 0.56

PI7: Therapist makes a clarification .62 636.13 1.11 0.21

PI8: Therapist makes a confrontation .61 431.11 1.13 0.34

PI9: Therapist admits his/her participation in rupture process .75 9.33 2.87 0.31

PI10: Therapist self-discloses countertransference feelings .74 14.33 2.89 0.24

PI11: Therapist explains or redefines tasks/goals of therapy .81 83.33 2.89 0.23

PI12: Therapist makes an interpretation .64 39.67 2.11 0.44

PI .67 996.37 1.00 0.08

Negative interventions

NI1: Therapist seems to impose his/her worldview or gives unwanted advice .61 4.33 1.25 0.45

NI2: Therapist seems to compete with patient .62 11.66 2.18 0.44

NI3: Therapist seems to press patient on specific topic .66 12.33 1.80 0.67

NI4: Therapist seems doubtful about strategies .61 8.11 1.25 0.91NI5: Therapist changes offhand topic .71 4.66 1.60 0.45

NI6: Therapist intellectualizes or is not focused on patient experience .70 14.66 1.46 0.45

NI7: Therapist talks in technical jargon .72 10.66 1.6 0.46

NI8: Therapist is hostile .68 14.33 2.66 0.35

NI .66 5.33 1.16 0.32

Note. P0mean presence; M0mean intensity scores; SD0mean SD of intensity scores.aMean kappa values of the three raters on a total of 2,984 patient utterances and 2,984 therapist utterances. Kappas have been calculated

considering only the agreement on the categories and not on the intensity score. Kappa values were rated as follows:B00poor; 0.200

slight; .21.400fair; .41.600moderate; .61.810substantial; .811.000almost perfect (Landis & Koch, 1977).


8/2/2019 CIS AC-VL 2009

9/18

for collaboration processes: '10low collaboration;

'20mid collaboration; '30high collaboration).

The mean intensity scores are called, respectively,

patient collaboration level (PCL) and therapist

collaboration level (TCL).

Every patient utterance is evaluated in considera-

tion of both the previous therapist intervention and

the previous patient

therapist interactions. To rate apatient utterance, the coder first has to look for the

broader category that is occurring (DRM, IRM, or

CP). Once the category is identified and coded, the

rater looks for the specific occurring behavior (within

the broader category that is being rated). Once a

category is coded, all the other behaviors automati-

cally become coded as not occurring. If the specific

patient behavior that is occurring is not clearly

recognizable, the rater assesses only the presence of

the broader category. First, the rater has to check for

a DRM; if not present, the rater will consider an

IRM and only at the end a CP. We tested several

rating algorithms over the years, and this one provedto be the fastest and most reliable rating procedure.

After the utterance is coded, raters assess the level

of collaboration. If the utterance has been coded as a

rupture, the rater will assign a negative evaluation,

and if coded as a resolution, a positive intensity

score. We suggest the following assignations: 3

when the patient is verbally attacking the therapist or

the therapy, showing distress in the relationship; 2

when the patient adopts a passive/aggressive way to

manifest hostility, for example, using sarcasm or

adopting avoidance measures in response to thera-

pist attempts to explore problems in the relationship;1 when the patient partially disengages from the

psychotherapy process or relationship through mini-

mizing strategies, such as intellectualization, avoid-

ance measures, or shifting from relevant topics to

more peripheral ideas; '3 when the patient relates a

topic to a topic or a schema to others situations,

discusses problems in the relationships in a colla-

borative and reflexive fashion, onnects problems in

the therapeutic relationship with other significant

relationships; '2 when the patient clarifies the

intensity or quality of his or her feelings, discloses

feelings in relation to a problem in the relationship,

or brings a new significant topic related to a therapist

exploration of problems in the relationship; '1

when the patient clarifies or reports significant new

topics, explores several elements regarding a signifi-

cant issue, or introduces elements within a topic.The evaluation of each therapist intervention is

made in consideration of both the previous patient

utterance and the previous patienttherapist interac-

tions. To rate a therapist utterance, the coder must

first look for the broader category that is occurring

(NI or PI). The main criteria for the evaluation of

the quality of an intervention (negative vs. positive)

are (a) emotional content (e.g., aggressive vs colla-

borative, as when a therapist responds in a comple-

mentary manner to patient hostility); (b) focus on

the concrete experience of the patient (vs. intellec-

tualization, as when the therapist minimizes the

emotional content of an interaction to avoid overlyintense feelings; (c) clarity (vs. vague, incomprehen-

sible interventions, as when the therapist is pressed

by the patient and this affects the clarity of the

intervention and its formulation).

Once the category is identified and coded, the

rater looks for the specific occurring intervention

(within the broader category that is being rated). If

the specific therapist intervention that is occurring is

not clearly recognizable, the rater assesses only the

presence of the broader category. First, the rater has

to check for an NI; if not present, the utterance will

be automatically rated as a PI.After the utterance is coded, the rater must rate

the level of collaboration of the therapist interven-

tion. We suggest the following assignations: 3 when

the therapist clearly manifests hostility toward the

patient, is sarcastic or devaluating; 2 when the

therapist avoids exploring patient negative feelings

(e.g.,, hostility or inappropriate erotization) if they

seem to be directed toward him or her or disengages

through intellectualization during rupture processes;

Table II. Different Patient Answers/Codings to the Same Therapist Intervention

Rating Therapist: How was going back home? Comment

DRM What do you mean? (Screaming) I think it is not relevant to speak

about it! My problems are the panic attacks not my parents! Who cares

about going back home?

Patient strongly refuses therapists intervention and

seems to disagree with therapist on what is important to

explore for an understanding of his problems.

IRM Going back home? Hmm, yes . . . hmm. Well, life is made of going away

and coming back! Yesterday I went to the new music store near the train

station. Have you been there?

Patient withdraws from internal states and tries to

intellectualize about his experience and to shift the focus

of the communication.

CP Well, it was not easy to see my mother. I was very nervous and at thesame time rather absent-minded . . . every time I see my mother I hope

something can change, to see her happier.

Patient speaks about his feelings and tries to give anemotional shade to the discourse.

Note. DRM0direct rupture marker; IRM0indirect rupture marker; CP0patient collaborative process.


8/2/2019 CIS AC-VL 2009

10/18

1 when the therapist partially disengages from the

psychotherapy process or relationship through in-

tellectualization of patients suffering or communi-

cates in an insufficiently clear way; '3 when the

therapist focuses attention on the here and now of

the relationship, responds in a collaborative manner

to patient hostility, stimulates a sense of we-ness; '2

when the therapist explores different patients states

and feelings about the relationship during a resolu-tion process, clarifies patients experience, or works

on patients emotions after the process of resolution;

'1 when therapist interventions are not explicitly

addressed to therapeutic alliance but nonetheless

contribute in a positive way to the psychotherapy

process.

Codes are mutually exclusive. A patient utterance

cannot be rated both as a rupture (DRM or IRM)

and as a collaborative process (CP), nor can a

patient rupture be rated at the same time as direct

or indirect. In the same way, a therapist intervention

cannot be rated as both negative and positive. If a

patient sentence contains both elements of colla-boration and rupture*in our experience a not very

common event*the rater is asked to make a choice,

considering the relevance of the two aspects in the

sentence, their sequence, and, if necessary, the

intention of the previous and the following utter-

ances. Rarely, in different parts of long sentences or

in very animated interactions, it is possible to detect

two categories of the same type (e.g., DRM1 and

DRM4); in these cases, we suggest rating both

categories. In this study, however, to stress scale

reliability for each single item, we asked raters to

code only the category that was most representativeof the sentence.

The time needed for the rating process of a 45- to

50-min transcript depends on the complexity of the

case and rater experience. The procedure (including

a preliminary reading of the transcript) takes

approximately 2 hr per session transcript. A junior

rater may need 3 hr for a complete evaluation, while

a senior rater (with at least 3040 rated sessions with

CIS) may require about 2 hr.

Qualitative and Quantitative Analysis. The CIS

furnishes a great deal of information about the way

patient and therapist construct their collaboration.

Once the transcript is rated, we have information foreach therapistpatient interaction about the intensity

of the collaboration of both patient and therapist

(PCL and TCL) as well as information about the

typologies of collaboration (patient and therapist use

of specific rupture markers and interventions).

The analysis of patient and therapist moment-by-

moment collaboration levels furnishes information

about the intensity of the collaboration (negative

from 1 to 3; positive from '1 to '3), the trend of

the collaboration during the session, and its positive

or negative fluctuations. A graphic representation

(Figure I) is useful for an immediate understanding

of PCL and TCL.

The scale also furnishes information about the

characteristics of the collaboration: the use by the

patient of specific typologies of rupture or a parti-

cular therapist style of intervention, the occurrence

of specific rupture markers in response to therapist

intervention, and so on. This kind of data can be

analyzed in different ways, but dynamic factor

analysis and sequential analysis seem to be the

most useful because these methodologies provide

the means to describe the idiosyncratic interaction of

every patienttherapist dyad.

Once the evaluation is completed, it is alsopossible to divide the session into several main

rupture and collaborative interactions. Rupture in-

teractions are characterized by the presence at least

of one patient rupture marker or one therapist

negative intervention. Collaborative interactions are

Patient Therapist Collaboration Level

3

2

1

0

1

2

3

1 8 15 22 29 36 43 50 57 64 71 78

Collboration

Level

PCL

TCL

TherapistRupture

Therapist/PatientCollaborations

PatientRupture

TherapistRupture

Figure I. Graphic representation of patient and therapist collaboration level for each interaction during a session. PCL0patient collabora-

tion level for each therapistpatient interaction; TCL0therapist collaboration level for each therapistpatient interaction; therapist

rupture0interaction in which a therapist negative intervention is present; patient rupture0interaction in which a patient rupture marker

(direct or indirect) is present; y-axis: therapistpatient interaction; x-axis: rupture (3, 2, 1) and collaboration ('1, '2, '3) intensity

scores.


8/2/2019 CIS AC-VL 2009

11/18

characterized by the presence at least of one patient

collaborative process and one therapist positive

intervention (see Figure I for graphic exemplifica-

tions). The consecutive occurrence of these kind of

interactions determines the presence, respectively, of

rupture phases (with more than one rupture inter-

action) and collaborative phases (with more than one

collaborative interaction). The two phases may vary

in both duration (from small phases of two interac-tions to larger phases of many interactions) and

intensity (from subtle shifts to more significant

fluctuations in the collaboration level). They can

also alternate during the session; some sessions may

present a rapid, ongoing alternation of rupture and

collaborative interactions that are not organized in

phases.

Using a quantitative approach, it is also possible to

select session interactions in which the collaboration

level (patient, therapist, or both) shows a significant

positive or negative shift (e.g., a 1.5 SD) and to

analyze patient and therapist interaction sequencesthat precede these shifts.

Although the CIS has been developed to assess

collaboration fluctuations within sessions, we also

developed global indexes that can be helpful for a

quantitative analysis of change over time (e.g.,

comparing different phases of a treatment): Index

of Direct Ruptures, Index of Indirect Ruptures,

Index of Collaborative Processes, Index of Negative

Interventions, and Index of Positive Interventions

(IPI; see Table III). All these indexes take into

consideration the mean frequency (in relation to

the number of therapistpatient interactions) and themean intensity (mean of the intensities rated for the

specific categories in the whole session) of a specific

category (DRM, IRM, PI, and NI). These formulas,

based on the mean frequency and intensity of

specific categories within the session, can inform us

about the main characteristics of the session (e.g.,

patient and therapist rupture style, global patient

and therapist collaboration).

Coding Manual. The coding manual explains the

scale structure and the rating procedure. Coding

rules, at least two clinical examples, and the ratingdifferential criteria are presented for each item. In

drafting this manual, we took inspiration from

a number of other manuals. For example, the

Table III. Global indexes for Quantitative Analysis

Index Formula Comment

IDR S DRM/S TP interactions) mean

intensity of DRM)10

This formula is based on the mean frequency of DRM in a session (total

number of DRM/total number of interactions) multiplied for the mean

intensity of DRM. The result must be a multiple of 10 to have a more useful

and practice range of scores and to increase variance. The scores range from

030.

IIR S IRM/S TP interactions) mean

intensity of IRM)10

This formula is based on the mean frequency of IRM in a session (total

number of IRM/total number of interactions) multiplied for the mean

intensity of IRM. The result must be a multiple of 10 to have a more useful


030.

ICP S CP/S TP interactions) mean

intensity of CP)10

This formula is based on the mean frequency of CP in a session (total

number of DRM/total number of interactions) multiplied for the mean

intensity of CP. The result must be a multiple of 10 to have a more useful


030.

IPI S PI/S TP interactions) mean

intensity of PI)10

This formula is based on the mean frequency of PI in a session (total

number of PI/total number of TP interactions) multiplied for the mean

intensity of PI. The result must be a multiple of 10 to have a more useful


030.

INI S NI/S TP interactions) mean

intensity of NI)10

This formula is based on the mean frequency of NI in a session (total

number of NI/total number of TP interactions) multiplied for the meanintensity of NI. The result must be a multiple of 10 to have a more useful


030.

Note. IDR0Index of Direct Ruptures; IIR0Index of Indirect Ruptures; ICP0Index of Collaborative Processes; IPI0Index of Positive

Interventions; INI: Index of Negative Interventions. S DRM: total number of direct rupture markers in the whole session; S TP

interactions0total number of therapistpatient interactions; S IRM0total number of indirect rupture markers in the whole session; S

CP0total number of collaborative processes in the whole session; S PI0total number of positive interventions in the whole session; S NI0

total number of negative interventions in the whole session; mean intensity0mean intensity of each category (DRM, IRM, PI, NI) is the

result of the sum of the intensity scores assigned to the category in the whole session divided by the number of occurrences of the category.


8/2/2019 CIS AC-VL 2009

12/18

formulation of the rating criteria for IRM4 (Patient

denies an evident feeling state) and IRM5 (Patient

intellectualizes about his/her inner experience) was

inspired by the Defense Mechanism Rating Scales

coding manual (Perry, 1990). The rating procedure

for IRM6 (Patient alludes to negative sentiments or

concerns about the therapeutic relationship through a

thematically linked discussion of out-of-session events or

relationships) was partially derived from the PatientExperience of the Relationship with the Therapist

measure (Gill & Hoffman, 1982). The reflective

functioning coding manual (Fonagy, Target, Steele,

& Steele, 1998) helped us to clarify the distinction

between a high collaboration process of the patient

(CP3) and a sophisticated intellectualization or use

of mental states as cliche. Some therapist interven-

tion rating criteria were also derived from the

Psychodynamic Interventions Rating Scale (Cooper

& Bond, 1992).

A large part of the coding manual is dedicated to

training exercises. These are presented according to

an increasing level of difficulty, from short sentencesto whole sessions. Our manual also includes three

sessions rated and commented on by the authors.

Method

Patients. We evaluated 32 session transcripts

(2,984 patient utterances and 2,984 therapist utter-

ances) of 16 patients (six men, 10 women; mean

age029.91 years, SD010.12) in psychotherapy.

Patients and sessions were chosen randomly from

our database. First, we randomly selected the cases

and then randomly selected two sessions for each

case. All patients were recruited from private prac-tice. We did not have information about outcome

(outcome evaluations were not performed or avail-

able). Before entering psychotherapy, all patients

received a Diagnostic and Statistical Manual of Mental

Disorders (fourth edition, text revision; American

Psychiatric Association, 2000) diagnosis. Ten

patients had at least one Axis II diagnosis: four met

criteria for borderline personality disorder, two for

borderline personality disorder and dependent

personality disorder, one for histrionic personality

disorder, and three for narcissistic personality dis-

order with dysthymic features. Six patients had only

one Axis I diagnosis (three with dysthymia; two with

obsessivecompulsive disorder; one with major

depression). All patients signed informed consent

to participate in the research.

Therapists. Sixteen therapists (seven cognitive,

nine psychodynamic), each with at least 10 years of

clinical experience, participated in the study. The 10

men and six women were a mean age of 44 years

(SD08.5). All the psychotherapies were adminis-

tered once a week and lasted a mean of 95 sessions

(range: 81156, SD023). The mean number of

coded session number was 50.71 (SD045.57,

range: 1123).

Raters. Three raters blind to study conditions

rated sessions using the CIS. Raters were postgrad-

uate clinical psychology students in the third year of

clinical training and specifically trained in using the

CIS. Training consisted in studying the codingmanual, participating in group discussions and

consensus rating of 10 session transcripts, as well

as homework. Classroom training lasted 16 hr,

including 2-hr sessions twice a month, in addition

to the homework. Sessions used for training were not

included in the study. To partially reduce some rater

biases (e.g., preferences for a specific psychother-

apeutic approach, expectations of a better or worse

session in relation to the diagnosis or phase of the

therapy), we gave no information about therapies

and diagnoses.

Results

Occurrence and Intensity of Ruptures and

Collaborative Processes. Mean frequency and intensity

for each category (DRM, IRM, CP, PI, NI) were

calculated. Means were calculated by averaging the

ratings of the three raters across the 32 sessions.

Table IV shows the frequency and mean intensity for

each category.

Patient CPs and therapist PIs were the most

frequently rated. One hundred percent of the

sessions contained at least one IRM compared with

43% for DRMs. NIs were rated in 31% of the

sessions. The intensity rated for DRMs was greater

than for IRMs, whereas mean intensity ratings for NI

were lower than for PI.

Interrater Reliability. Overall interrater agreement

between the three raters for PCL and TCL was

Table IV. Mean Occurrence and Intensity for Session of Patient

and Therapist Ruptures and Collaborative Processes

Presence Intensity

Rating M SD M SD

DRM 1.86 3.15 2.75 0.47

IRM 17.79 8.55 1.25 0.38

CP 73.37 9.23 1.11 0.49

PI 90.72 9.03 1.93 0.32

NI 2.69 1.47 1.66 0.51

Note. Mean presence and intensity have been evaluated calculating

the mean presence and intensity rated by every rater for every

session (n032). DRM0direct rupture marker; IRM0indirect

rupture marker; CP0collaborative process; PI0positive inter-

vention; NI0negative intervention.


8/2/2019 CIS AC-VL 2009

13/18

evaluated. The data matrix was formed by six

columns: two for every rater (one for PCL and one

for TCL). Rows represented therapistpatient inter-

actions. Interrater reliability was evaluated in 2,984

patient utterances and 2,984 therapist utterances. To

evaluate the agreement on collaboration level, we

used the intraclass correlation coefficient (Shrout &

Fleiss, 1979). The mean overall interrater agreement

of the three raters was 0.73 (range: 0.680.78): 0.76for PCL (range: 0.730.78) and 0.68 for TCL

(range: 0.680.70).

We were also interested in studying interrater

reliability for each CIS category (DRM, IRM, CP,

PI, NI), taking into consideration only the presence

of the evaluation, not its intensity. Interrater relia-

bility in this case was evaluated in 2,984 utterances

(2,984 for patient and 2,984 for therapist) using

Cohens kappa (Fleiss, 1981). The analysis was

performed on a classical confusion matrix, using

one for patient ratings and one for therapist ratings.

Kappa values were given the following ratings:B00

poor; 0.200slight; .21.400fair; .41.600mod-erate; .61 .810substantial; .811.000almost per-

fect (Landis & Koch, 1977). The average interrater

agreement was rated as almost perfect for the Direct

Rupture Marker scale (.81) and as substantial for

both the Indirect Rupture Marker scale (.66) and the

Collaborative Process scale (.72). Interrater reliabil-

ity was also substantial for Positive Intervention and

Negative Intervention scales (.67 and .66, respec-

tively). Detailed reliability estimates are presented in

Table I.

The Relationship between Therapist and PatientCollaboration. We evaluated the relationship between

therapist intervention (positive and negative) and

patient rupture markers and CPs (Table V). Results

indicated a positive correlation between therapist PIs

and patient CPs as well as between therapist NIs and

patient ruptures (direct and indirect). These are

correlational data, however, and at this level of

analysis we are unable to indicate the causal direc-

tion of correlations.

To calculate correlations, we used a data matrix in

which each row represented a therapistpatient

interaction and each column a CIS item (e.g., NI1,

NI2, DRM1, DRM3). Each cell contained a number

indicating the intensity of each single item (ranging

from 0 [not coded] to 3). In this way, we simulta-neously considered the presence and intensity of

each rating. For easier interpretation of the results,

we changed the sign of the intensity scores of all

patient ruptures and therapist negative interventions

(-3 to '3, 2 to'2, 1 to '1).

PIs were positively correlated with CPs, whereas

NIs were positively correlated with DRMs and

IRMs. NIs also correlated negatively with CPs.

Furthermore, results indicated a positive correlation

between therapist PI and patient IRMs. For a deeper

understanding of this correlation, we calculated the

correlations between all the typologies of therapist

PIs and patient DRM, IRM, and CP (Table VI).As shown in Table VI, some therapist PIs corre-

lated exclusively with patient CPs, whereas other

therapist PIs correlated with patient CPs and DRM

and/or IRM. Expressive interventions such as PI8

(Therapist makes a confrontation) and PI12 (Therapist

makes an interpretation) were more correlated with

IRM than with CP. Moreover, PI11 (Therapist

explains or redefines tasks/goals of therapy) appeared

highly correlated with patient DRM.

Discussion

In this study, we (a) presented the CIS for the first

time, (b) assessed the occurrence of patient ruptures

and CPs and therapist PIs and NIs, (c) tested scale

reliability, and (d) evaluated the relationship among

therapist interventions, patient rupture markers, and

CPs. In all sessions, at least one IRM was present

whereas DRMs did not always appear. The mean

presence per session of IRM was greater than for

DRM. This result is quite similar to that of another

study investigating the occurrence of withdrawal and

confrontation ruptures in session transcripts (Som-

merfeld, Orbach, Zim, & Mikulincer, 2008). The

mean intensity score of the Direct Rupture Markersscale was higher than that of Indirect Rupture

Markers: DRMs are stronger strains in the alliance

than IRMs. Finally, patient CPs and therapist PIs

were the most frequent categories. The study did

not aim to analyze the relationship between CIS

ratings and other fundamental variables such as

therapy outcome and patient personality organiza-

tions. Nevertheless, these are crucial issues for

future research investigating the relationship among

Table V. Pearson Correlations between Therapist Positive and

Negative Interventions and Patient Rupture Markers and Colla-

borative Processes

Variable DRM IRM CP

PI .217 .512* .676*NI .781* .642* (.621*

Note. N02,984 utterances. Correlations have been calculated

using Person correlation coefficient. We evaluated the correlations

considering the intensity score (from 03) for each item of the

scale. We reversed the intensity scores of all patient ruptures and

therapist negative interventions (1, 2, 3 in '1, '2, '3).

DRM0direct rupture marker; IRM0indirect rupture marker;

CP0patient collaborative process; PI0positive intervention;

NI0negative intervention.

*pB.05.


8/2/2019 CIS AC-VL 2009

14/18

outcome, therapeutic approach, diagnosis and rup-

tures frequency, intensity, and so on.Overall reliability was good and in line with the

findings of similar measures, such as the Client

Resistance Code (Watson & McMullen, 2005)

and the Psychodynamic Intervention Rating Scale

(Milbrath, Bond, Cooper, Znoj, Horowitz, & Perry,

1999). Some items proved to be less reliable than

others, which could be partially due to the raters

lack of clinical experience. Another factor affecting

the reliability of these items may be a particular form

of disagreement, known as the location disagree-

ment (Luborsky et al., 1983). Some items, for

example IRM7 (Patient interacts in an acquiescent

manner), refer more to an interpersonal atmosphere

than an actual interaction. Therefore, it could be

difficult to recognize the beginning and the end of

this marker. Checking the single ratings and focusing

on IRM7, we realized that the three raters rated

IRM7 more or less the same number of times

(frequency) but in different verbal units of the whole

session.

Therapist NIs and patient rupture markers

(DRMs and IRMs) were positively correlated. NIs

seem to be produced by therapists in very difficult

moments of the session, a correlation that may

indicate that the therapist is responding in a defen-sive, noncollaborative way to patient ruptures. (Keep

in mind that we evaluate therapist interventions as

the therapists response to the previous patient

utterances.) At the same time, this association could

be explained as the effect of these negative interven-

tions on patient response: From this point of view,

DRM and IRM could be seen as a product of

therapist NIs. At this level of analysis, we are unable

to indicate the causal direction of correlations, that

is, if the therapist NI provokes a patient rupture

marker or if it is a consequence of a rupture marker.A circular dynamic is the most probable solution.

This correlation could also be affected by rater

expectations insofar as raters are more likely to see a

rupture after a negative therapist intervention. From

a methodological point of view, this issue could be

addressed by using separate raters for patient and

therapist utterances (one rater blindly evaluates only

patient utterances and another blindly evaluates only

therapist utterances). Our coding system is based on

the concept of therapeutic alliance, however, and

thus it has been conceived for the evaluation of

single (patient or therapist) utterances in answer to

the previous interactions; it could be misleading to

assess the two contributions separately. Possible

effects of rater expectations have to be taken into

consideration, but we realized that good observation

skills united with adherence to the coding manual

guidelines can prevent this effect.

Therapist PIs are significantly correlated with

patient CPs. This correlation probably describes

those moments of the therapy in which patient and

therapist work together in a collaborative atmo-

sphere: Therapist responds in a collaborative way

to patient collaboration, which positively affects the

subsequent patient response.PIs were also correlated with some IRMs and, to a

lesser degree, with some DRMs. Different explana-

tions can be given for this phenomenon. For example,

these correlations may describe therapistpatient

interactions in which the therapist responds to a

previous patient rupture communication in a colla-

borative way. Another explanation could be that these

interventions may provoke a negative reaction in the

patient instead of the expected positive one. As

Table VI. Correlations between Therapist Positive Interventions and Patient Direct Rupture Markers, Indirect Rupture Markers, and

Collaborative Processes

Therapist Intervention DRM IRM CP

PI1: Therapist focuses on the here and now of the relationship .125 .512* .653*

PI2: Therapist explores different patient states .096 .571* .715*

PI3: Therapist provides feedback to patient .111 .531* .712*

PI4: Therapist suggests a patient emotion .096 .631* .688*

PI5: Therapist believes that patient is indirectly talking about

relationship

.649* .756* .145

PI6: Therapist furnishes an empathic sustain to patient .112 .145 .788*

PI7: Therapist makes a clarification .095 .212 .675*

PI8: Therapist makes a confrontation .288 .643* .518*

PI9: Therapist admits his/her participation in rupture process .078 .434 .618*

PI10: Therapist self-discloses countertransference feelings .122 .512* .545*

PI11: Therapist explains or redefines tasks/goals of therapy .698* .455* .325

PI12: Therapist makes an interpretation .143 .612* .532*

Note. To obtain a single correlation coefficient, we calculated the correlations between PI and DRM, IRM, and CP for every rating of the

three raters and then we calculated the overall average correlation. DRM0direct rupture marker; IRM0indirect rupture marker; CP0

patient collaborative process; PI0positive intervention.

*pB.05.


8/2/2019 CIS AC-VL 2009

15/18

already observed, at this level of analysis, we are

unable to indicate the causal direction of correlations,

in this case if the correlation represents a therapist

addressing ruptures or, conversely, a patient respond-

ing with a rupture to something we rated as a PI.

The fact that therapist PIs may provoke patient

negative answers is a critical aspect that needs

further investigation. In this study, we only calcu-

lated the moment-by-moment correlation: Probably,considering different lags (evaluating the effect of a

therapist intervention also in relation to a series of

antecedent and successive patient responses), we

could obtain more exhaustive information about this

aspect.

The aim of this study was to present and explain a

new measure and to test its reliability. Future

research must investigate other aspects, such as

construct and convergent validity. Another relevant

issue to test will be the relationship between in-

session ruptures and resolutions (as evaluated by

CIS) and sessions and therapies outcome. This is a

critical aspect for us in relation to not only thevalidity of our instrument but also clinical practice.

As clinicians and researchers, we need more infor-

mation about what to do and how to address

therapeutic impasses, particularly in the treatment

of severe personality disorders, which are character-

ized by rupture processes more so than other

pathologies.

Future research could further develop the Safran

and Muran model, on which our scale is largely

based. Some examples are a special focus on

therapists contributions in order to define more

detailed rupture models from the therapist perspec-

tive, studying specific therapist actions that nega-

tively or positively contribute to the alliance process,

studying the link between specific kinds of ruptures,

specific personality disorders, and so on.

The CIS represents an effort to develop a scale for

studying alliance ruptures and repair processes

starting from an analysis of transcripted in-session

interactions. This is in line with other research

designed to evaluate the occurrence of ruptures at

a turn-by-turn conversational level (Watson &

McMullen, 2005; Sommerfeld et al., 2008).

Other strategies to evaluate alliance rupture pro-

cesses are based on the task analytic investigationparadigm proposed by Greenberg (2007) and on the

idea that it is possible to identify and delineate a

rupture episode. This strategy could make it possible

to study some rupture episodes more in detail and

build process models for resolution ruptures (Asp-

land, Llewelyn, Hardy, Barkham, & Stiles, 2008;

Safran et al., 1994). As described in the Qualitative

and Quantitative Analysis section, CIS is not pre-

sently conceived for the assessment of rupture

episodes, even if we can divide the session into

rupture and collaborative phases once the evaluation

is completed.

Before concluding, we should point out that

evaluating transcripts requires a great deal of energy:

transcribing sessions, training raters, rating sessions.

It can be very onerous, so the application of our scale

is indicated for single-case designs or very small

samples.As we developed this scale, we also tested scoring

units larger than one utterance unit (e.g., 5 min,

one-sixth of a session). Adopting a larger scoring

unit has its advantages, including for example, a

reduction in rating time and a probable increase in

interrater reliability. On the other hand, it can result

in the loss of some information, especially that

related to the sequentiality of patient and therapist

actions. In conclusion, the use of different scoring

units depends on research aims. For studying

idiosyncratic interactional patterns, we believe it is

necessary to adopt a microanalytic level of analysis;

however, for studying the correlation between rup-ture episodes and outcome, it could be useful to

adopt a larger scoring unit.

Several years spent training residents and young

psychotherapists made us appreciate the utility of the

CIS as a training tool. In fact, it is a good way to help

trainees develop their ability to recognize subtle

rupture episodes.

At a more general level, the CIS represents an

effort to bridge the gap between clinicians and

researchers and to contribute to the development

of a clinically articulated but empirically grounded

way of assessing therapeutic alliance and rupture

repair processes. Indeed, several recent studieshave illustrated the importance of combining quali-

tative and quantitative methodologies in exploring

therapeutic relationships (see Bucci, 2005; Kachele

et al., 2006; Lingiardi, Shedler & Gazzillo, 2006;

Shedler & Westen, 2006). We believe this approach

can help the therapist to clarify and articulate what

happens during the session, which kinds of inter-

ventions are more effective with a specific patient,

and which kinds of rupture and resolution dynamics

are prevalent with that patient in relation to specific

moments or topics of the therapy. From this point of

view, the CIS is not only a taxonomy of therapeutic

alliance ruptures and resolutions but also a tool

aimed at increasing clinical knowledge of patient

therapist dynamics and interactions.

Acknowledgements

We thank all the junior and senior colleagues who, in

the course of the years, have been contributing to the

scale development: all the participants of our clinical


8/2/2019 CIS AC-VL 2009

16/18

seminars on therapeutic alliance assessment, in

particular the group from Sapienza University of

Rome (Francesco De Bei, Francesco Gazzillo, Da-

niela Gentile); Daniela Maggioni and the ASP

Clinical and Research Group of Milan; the group

from Padua University (Adriana Lis, Silvia Salcuni,

Diego Rocco); and the group from Bicocca

University of Milan (Marta Vigorelli, Mariangela

Villa, Tiziana Porta). In addition, we thank theanonymous reviewers for their stimulating com-

ments and suggestions.

References

Ackerman, S., & Hilsenroth, M. (2001). A review of therapist

characteristics and techniques negatively impacting the ther-

apeutic alliance. Psychotherapy, 38, 171185.

Ackerman, S., & Hilsenroth, M. (2003). A review of therapist

characteristics and techniques positively impacting the ther-

apeutic alliance. Clinical Psychology Review, 23, 133.

Alexander, L. B., & Luborsky, L. (1987). The Penn Helping

Alliance Scale. In L. S. Greenberg & W. M. Pinsof (Eds), The

psychotherapeutic process: A research handbook (pp. 325356).New York: Guilford Press.

Allen, J. G., Newsom, G. E., Gabbard, G. O., & Coyne, L.

(1984). Scales to assess the therapeutic alliance from a psycho-

analytic perspective. Bulletin of the Menninger Clinic, 48, 383

400.

American Psychiatric Association. (2000). Diagnostic and statistical

manual of mental disorders (4th ed.., text rev.). Washington, DC:

American Psychiatric Association.

Aspland, H., Llewelyn, S., Hardy, G. E., Barkham, M., & Stiles,

W. (2008). Alliance ruptures and rupture resolution in cogni-

tive-behavior therapy: A preliminary task analysis. Psychother-

apy Research, 18, 699710.

Beebe, B., & Lachmann, F. M. (2002). Infant research and adult

treatment. Hillsdale, NJ: Analytic Press.

Benjamin, L. S. (1984). Principles of prediction using structuralanalysis of social behavior (SASB). In R. A. Zucker, J. Aronoff

& A. J. Rabin (Eds), Personality and the prediction of behavior

(pp. 121173). New York: Academic Press.

Bordin, E. S. (1979). The generalizability of the psychoanalytic

concept of the working alliance. Psychotherapy: Theory,

Research, and Practice, 16, 252260.

Bordin, E. S. (1994). Theory and research in the therapeutic

working alliance: New directions. In A. O. Horvath & L. S.

Greenberg (Eds), The working alliance: Theory, research, and

practice (pp. 1337). New York: Wiley.

Bucci, W. (2005). Process research. In E. S. Person, A. M. Cooper

& G. O. Gabbard (Eds), The American Psychiatric Publishing

textbook of psychoanalysis (pp. 317333). Washington, DC:

American Psychiatric Publishing.

Charmann, D. (Ed.). (2004). Processes in brief psychodynamicpsychotherapy: Advancing effective practice. Hillsdale, NJ:

Erlbaum.

Colli, A., & Lingiardi, V. (2001, October). Una proposta di

valutazione dellalleanza terapeutica a partire dai trascritti delle

sedute: il sistema IVAT [A new transcript method for the

assessment of therapeutic alliance: The IVAT system]. Paper

presented at the national meeting of the Society for Psychother-

apy Research, Italian section, Palermo, Italy.

Colli, A., & Lingiardi, V. (2002). Indice di valutazione dellal-

leanza terapeutica [Therapeutic Alliance Evaluation Index]. In

V. Lingiardi (Ed.), LAlleanza terapeutica: Teoria, Ricerca, Clinica

[The therapeutic alliance: Theory, research, and practice] (pp.

212222). Milan: Raffaello Cortina.

Colli, A., & Lingiardi, V. (2003, November). Valutazione multi-

strumentale di una paziente in psicoterapia: La valutazione

dellalleanza terapeutica con lIVAT [A multistrumental assess-

ment of a single case: Evaluating therapeutic alliance with

IVAT]. Paper presented at the national meeting of the Society

for Psychotherapy Research, Milan, Italy.

Colli, A., & Lingiardi, V. (2005, September). Valutare i processi di

rottura e riparazione dellalleanza terapeutica: lIVAT-II [Assessing

alliance ruptures and resolutions with IVAT-II]. Paper pre-sented at the national meeting of the Society for Psychotherapy

Research, San Benedetto del Tronto, Italy.

Colli, A., & Lingiardi, V. (2006, June). Therapists interventions

adjustment ratio and relational failures: A pilot study on the way

therapists break the alliance. Paper presented at the 37th

International Meeting of the Society for Psychotherapy

Research, Edinburgh, Scotland.

Colli, A., & Lingiardi, V. (2007). Valutare lalleanza terapeutica

attraverso trascritti di sedute: lattendibilita della IVAT-R nello

studio di quattro psicoterapie [Assessing therapeutic alliance

with a transcript based method: Reliability of the IVAT revised

version in the study of four single cases]. Ricerca in Psicoterapia,

10, 7597.

Constantino, M. J., Marnell, M., Haile, A. J., Kanther-Sista, S.

N., Wolman, K., Zappert, L., et al. (2008). Integrativecognitive therapy for depression: A randomized pilot compar-

ison. Psychotherapy, 45, 122134.

Cooper, M., & Bond, S. (1992). Psychodynamic Interventions

Rating Scale. Unpublished manuscript.

Eaton, T., Abeles, N., & Gutfreund, M. J. (1993). Negative

indicators, therapeutic alliance, and therapy outcome. Psy-

chotherapy Research, 3, 115123.

Elkind, S. N. (1992). Resolving impasses in the therapeutic relation-

ships. New York: Guilford Press.

Elvins, R., & Green, J. (2008). The conceptualization and

measurement of therapeutic alliance: An empirical review.

Clinical Psychology Review, 28(7), 11671187.

Fleiss, J. L. (1981). Measuring nominal scale agreement among

many raters. Psychological Bulletin, 76, 378382.

Fonagy, P. Target, M. Steele, H. & Steele, M. (1998). Reflectivefunctioning coding manual. Unpublished manuscript, University

College, London.

Freud, S. (1958). The dynamics of transference. In J. Strachey

(Ed. and Trans.), The standard edition of the complete psycholo-

gical works (Vol. 12, pp. 312322). London: Hogarth Press &

Institute of Psychoanalysis. (Original work published 1912)

Gaston, L., & Marmar, C. R. (1994). The California Psychother-

apy Alliance Scale. In A. O. Horvath & L. S. Greenberg (Eds),

The working alliance: Theory, research and practice (pp. 85108).

New York: Wiley.

Gill, M. M., & Hoffman, I. Z. (1982). A metho

Date post:	05-Apr-2018
Category:	Documents
Upload:	alirezatahmasb
View:	214 times
Download:	0 times

CIS AC-VL 2009

Documents